The 1982-1983 Bangladesh nutritional blindness study visited 11,618 rural households and examined 18,660 preschool-age children in an effort to determine the prevalence and determinants of eye lesions and loss of sight due to vitamin A deficiency (xerophthalmia). Risk of xerophthalmia was significantly higher for children from households without any of the indicators of relative wealth used. Almost 80% of blind children from landless households, and even a very small garden reduced considerably the chances of a household having a xerophthalmic child. Poorer households with access to less than 0.3 acres land or no garden or without a tin roof, wristwatch, radio or cycle were at least twice as likely as their more fortunate neighbours to have a young child with any type of xerophthalmia. Taking account of such socio-environmental risk factor weightings would direct the scarce resources of intervention programmes to households and children who most need them.